Analysis: Taped vs Live RAW (2000-2013)
Since there's been the whole hullabaloo about whether being taped delayed affects the RAW rating in 2013, I decided to take a stab at the question.
I started with a dataset of RAW from 2000 to 2013 (about 693 weeks RAWs)..
This was the list of TAPED RAWs that I came up with:
taped........aired..........reason
12/22/2000 12/25/2000 HOLIDAY
12/29/2000 1/1/2001 HOLIDAY
12/21/2001 12/24/2001 HOLIDAY
12/21/2002 12/23/2002 HOLIDAY
10/11/2004 10/11/2004 ENGLAND (same day, tape delay)
2/4/2005 2/7/2005 JAPAN
4/25/2005 4/25/2005 ENGLAND (same day, tape delay)
11/13/2005 11/14/2005 TAPED (Ended up as Eddie Guerrero Tribute)
11/21/2005 11/21/2005 ENGLAND (same day, tape delay)
12/9/2005 12/19/2005 AFGHANISTAN (Tribute to the Troops 2005)
4/21/2006 4/24/2006 ENGLAND
10/15/2006 10/16/2006 LOS ANGELES (recorded a day earlier at the Staples Center)
11/13/2006 11/13/2006 ENGLAND (same day, tape delay)
12/8/2006 12/25/2006 IRAQ (Tribute to the Troops 2006)
2/12/2007 2/15/2007 PORTLAND (recorded several days earlier)
4/16/2007 4/16/2007 ITALY (same day, tape delay)
4/23/2007 4/23/2007 ENGLAND (same day, tape delay)
9/2/2007 9/3/2007 COLUMBUS (recorded a day earlier)
10/15/2007 10/15/2007 ENGLAND (same day, tape delay)
12/7/2007 12/24/2007 IRAQ (Tribute to the Troops 2007)
12/29/2007 12/31/2007 HOLIDAY
2/4/2008 2/11/2008 ONE WEEK EARLY (one live RAW; one recorded RAW)
4/14/2008 4/14/2008 ENGLAND (same day, tape delay)
11/10/2008 11/10/2008 ENGLAND (same day, tape delay)
4/20/2009 4/20/2009 ENGLAND (same day, tape delay)
6/29/2009 7/6/2009 SAN JOSE (one live RAW; one recorded RAW)
11/9/2009 11/9/2009 ENGLAND (same day, tape delay)
4/12/2010 4/12/2010 ENGLAND (same day, tape delay)
7/26/2010 8/2/2010 ONE WEEK EARLY (one live RAW; one recorded RAW)
8/16/2010 8/23/2010 ONE WEEK EARLY (one live RAW; one recorded RAW)
9/20/2010 9/27/2010 ONE WEEK EARLY (one live RAW; one recorded RAW)
11/8/2010 11/8/2010 ENGLAND (same day, tape delay)
4/18/2011 4/18/2011 ENGLAND (same day, tape delay)
6/27/2011 7/4/2011 ONE WEEK EARLY (one live RAW; one recorded RAW)
10/15/2011 10/17/2011 MEXICO (recorded days earlier)
4/16/2012 4/16/2012 ENGLAND (same day, tape delay)
11/5/2012 11/5/2012 ENGLAND (same day, tape delay)
12/18/2012 12/24/2012 HOLIDAY
12/29/2012 12/31/2012 HOLIDAY
4/22/2013 4/22/2013 ENGLAND (same day, tape delay)
Things that jump out:
- Since 2000, a quarter of all taped RAWs were taped in December
- 43% of the taped RAWs aired the same day, just on tape delay
- Between 2000 and 2007, the only RAWs that aired more than 3 days after they taped were Tribute for the Troops shows; meanwhile, there were six times between 2008 and 2012 that an episode of RAW aired a week after it was originally taped.
I ran a regression on "RAW rating" a function of several variables:
* 8-week trailing RAW rating average
* was there show taped? (1 or 0)
* was it a 3-hour RAW? (1 or 0)
* was it a the first RAW following a PPV? (1 or 0)
This simple regression would suggest a predictive formula (adjusted r-squared around 82%) of:
2000 to 2013 RAW Rating = 0.200 + 0.938x(8 week trailing RAW average rating) - 0.238x(was the show taped?) -.031x(is it a 3-hour RAW?) + 0.136x(is it a post-PPV RAW?)
Interestingly, the coefficient that has the highest p-value (making it statistically questionable) is the 3-hour RAW. Essentially, because 3-hour RAWs are now the norm, the meaningfulness of this variable is basically encapsulated in the trailing 8-week RAW rating average. (Or the 49 three-hour RAWs, only nine weren't part of this current consecutive run: 6/11/2012, 4/23/2012, 12/12/2011, 11/14/2011, 6/20/2011, 6/13/2011, 11/29/2010, 4/26/2010 and 10/3/2005).
So, this initial regression would suggest that a "taped" show impacts the show's rating negatively by about 0.24, which is a lot.
However, as I mention in the beginning - during 2000-2007, a taped RAW was mostly a Holiday RAW (in December) including Tribute to Troops shows, same day taped delay shows or aired within 3 days. I think the Holiday/December dates are tanking this calculation.
This leads to this inquiry paths: (1) Is there a more recent (2008-2013) trend around RAW ratings for tape delayed shows? (2) Does # of days since the tape delay affect the rating?
(1) What's the trend since 2008?
2008 to 2013 RAW Rating = 0.923 + 71.6%x(8 week trailing RAW average rating) - 0.180x(was the show taped?) -.096x(is it a 3-hour RAW?) + 0.126x(is it a post-PPV RAW?)
(All variables have a p-value of 0.03 or lower)
(2) Does # of days since the tape delay affect the rating?
2000 to 2013 Rating = 0.200 + 93.8%x(8 week trailing average) -0.10x(is the show taped?) -0.044x(how many days since it was taped?) + 0.138x(first RAW after a PPV?) -0.032x(is it a 3-hour RAW?)
(Taped and 3-hour raw have high p-values in this one; taped is 0.15 and 3-hours is 0.55.)
2008 to 2013 Rating = 0.94 + 71.2%x(8 week trailing average) -0.14x(is the show taped?) -0.015x(how many days since it was taped?) + 0.126x(first RAW after a PPV?) -0.101x(is it a 3-hour RAW?)
(Days have a high p-value in this one of 0.39)
Conclusion: So, this would suggest that same-day taped delay shows in 2013 (i.e. from the UK) would be impacted by around a tenth of a rating point drop beyond expected Rating.
The model estimates that RAW should have been a 3.03 and it did a 3.1. (If it wasn't taped it "would" have been 3.17 supposedly.)
I don't believe that a tenth of a rating point is worth fretting about since talent appearance move the needle a lot more.
Q: Do the numbers change much if you look at 2008 onward and remove the holiday shows?
A: Not a lot.
I added a Holiday Variable which was triggered for the Christmas Eve, Christmas Day, New Year's Eve or New Year's Day, July 4th and Memorial Day.
Projected RAW Rating = 0.99 - (0.14)x(Was the show taped?) - 70%x(8 week Trailing RAW Average Rating) - (0.10)x(Is it a 3 hour RAW?) - (0.33)x(Is it a Holiday show?) + 0.12x(Is it the day after a PPV?)
(p-value for all variables was 0.02 or lower. Adjusted r-squared is only 0.46. Days variable wasn't statistically significant enough to keep in the model.)
So, this model suggest that taped shows do have a slightly ratings decrease (again, around -0.1 ratings point). However, the lowered r-squared (down to 46%) suggests to me that there is a lot of week-to-week ratings variation that isn't being explained by the variables - i.e. big character returns or other factors (televised competition) contribute a lot to whether you're gaining or losing viewers. The quarter-hour trends lend a lot to this because we've seen how certain people can hold an audience while others have shaky/consistent-loss track records.
In short, methinks people like to watch The Rock.
-Chris Harrington
Post-script: Dave Meltzer noted that, "WWE actually did the exact same study internally last week. One thing noted is that the special shows that are expected to do better, day after Mania, day after Summer Slam, day after Rumble, appearance by major stars coming back, theme shows, are virtually always live. If you factor that out, and throw out the holiday shows which are going to do lower numbers, the difference between live and tape is very little."